I think part of the challenge in this element of game design is that the setting books & games establish that a Source's magic is barely controlled and strong stressors / negative emotion can unleash it. They are regarded as dangerous by many mages because of this. I just probably went too far in the dangerous direction with my design. So scaling the surges back but still making them meaningful/scary displays of The Power is a challenge.
Yar, it definitely is. It's a danger of working from an established body of fiction as well...a book doesn't need to worry about game balance or group dynamics. The author has utter freedom to determine exactly what happens and when, with only the consequences they decide. A game complicates that situation considerably by adding in player agency and a lot of random happenstance.
I say this as someone who has fallen afoul of it several times before, both as player and creator.
Another issue you pointed out is that "emotion as a trigger" is inherently nebulous. I don't really see a way around this one, it's just something that takes DM-player cooperation/trust to find pacing that "feels right." The whole "provide 3 situations" thing I came up with just to provide SOME sort of guideline. I'm fine with that if you are. I agree that toning down the impact of the wild surges will make the lack of rules for how common surges occur less of an issue.
There are ways to do it, but probably not within the framework of 5e without introducing a lot of unwieldy house rules. In preference to that a narrative approach like you have can work decently well I think...but some guidance for both players and GM's would not come amiss. And, as you point out, scaling the effects back will help mitigate the core issue.
I'm open to ideas if you see a better way of handling the whole "barely controlled magic" trope?
First of all, I'd say lets dispense with the big emotional category blocks. Having all three triggers stem from a single emotion means that it's much more likely that a Surge would logically trigger whenever that emotion comes up. There's only so many things to be angry or afraid about, after all. Unless the triggers are very specific in scope, having three separate triggers within a narrow emotional bandwidth makes them harder to avoid...both for players and GMs.
But, if I have the flexibility to say, "Talashia gets MOST angry about this one thing, and that's one thing that can trigger her. She's also SUPER guilty about this other thing, and that can trigger her...and she has a raving phobia of canned beans, so NEVER EAT CANNED BEANS AROUND HER. That lets me focus ONLY on the things that are the most intense for her. Wheras if I have to choose three things she gets insanely angry over...that's a bit of a pickle.
Secondly, give the player some agency over the surge. For example, perhaps a Source who feels a surge coming on can spend all their actions for the turn trying to resist it...make a Wisdom save or Charisma save or something at their own spellcasting DC (the stronger the magic, the harder to control!). This lets them say, "Oh no! It's happening! RUN!" They can keep doing it, but eventually the roll will come up bad and BOOM. The surge doesn't stop, but they can give their pals a chance to get clear.
Thirdly, use more than a d6 roll. There is not a lot of randomness in a d6 roll, for something like this. I know it's more work to set up, and certainly doing multiple 1d100 lists is out of the question, but put some more variability in there. That also lets you do things like have more extreme options weighted less...so maybe a lot of options on the list have 2 or 3 chances in 20 to come up, but the Cosmic Devourer Pokes His Proboscis Through the Multiverse and Nibbles your Cape only has a 1 in 20 chance. That makes it more rare, more special, and you can feel a little freer to give it consequences.
Fourth, define things. Example - Right now the result where a gate to another world appears and sucks people into it? We don't know how to determine where that gate appears. It's pretty crucial information to have. I assume the GM decides, but if so then say it because otherwise there'll be some bad blood.
Fifth, you might want to scale some of the effects with caster level. 4d10 is an incredibly huge amount of damage at 1st level. By 10th it's kind of meh. At 20 you barely notice it. Again, more power = more risk. When a level 20 Source blows his stack, people should probably be looking around for Mount Arrarat, because it was there when they left the cabin this morning. Conversely, a level 1 Source going nuclear should move a commensurately smaller molehill.
Not to say that a Surge is ever 'ignorable,' but scaling effects will really reduce the potential for one bad roll to cause a wipe...and that's something virtually guaranteed to create an OOC problem at any table.
Yar, it definitely is. It's a danger of working from an established body of fiction as well...a book doesn't need to worry about game balance or group dynamics. The author has utter freedom to determine exactly what happens and when, with only the consequences they decide. A game complicates that situation considerably by adding in player agency and a lot of random happenstance.

Another issue you pointed out is that "emotion as a trigger" is inherently nebulous. I don't really see a way around this one, it's just something that takes DM-player cooperation/trust to find pacing that "feels right." The whole "provide 3 situations" thing I came up with just to provide SOME sort of guideline. I'm fine with that if you are. I agree that toning down the impact of the wild surges will make the lack of rules for how common surges occur less of an issue.
There are ways to do it, but probably not within the framework of 5e without introducing a lot of unwieldy house rules. In preference to that a narrative approach like you have can work decently well I think...but some guidance for both players and GM's would not come amiss. And, as you point out, scaling the effects back will help mitigate the core issue.
I'm open to ideas if you see a better way of handling the whole "barely controlled magic" trope?
First of all, I'd say lets dispense with the big emotional category blocks. Having all three triggers stem from a single emotion means that it's much more likely that a Surge would logically trigger whenever that emotion comes up. There's only so many things to be angry or afraid about, after all. Unless the triggers are very specific in scope, having three separate triggers within a narrow emotional bandwidth makes them harder to avoid...both for players and GMs.
But, if I have the flexibility to say, "Talashia gets MOST angry about this one thing, and that's one thing that can trigger her. She's also SUPER guilty about this other thing, and that can trigger her...and she has a raving phobia of canned beans, so NEVER EAT CANNED BEANS AROUND HER. That lets me focus ONLY on the things that are the most intense for her. Wheras if I have to choose three things she gets insanely angry over...that's a bit of a pickle.
Secondly, give the player some agency over the surge. For example, perhaps a Source who feels a surge coming on can spend all their actions for the turn trying to resist it...make a Wisdom save or Charisma save or something at their own spellcasting DC (the stronger the magic, the harder to control!). This lets them say, "Oh no! It's happening! RUN!" They can keep doing it, but eventually the roll will come up bad and BOOM. The surge doesn't stop, but they can give their pals a chance to get clear.
Thirdly, use more than a d6 roll. There is not a lot of randomness in a d6 roll, for something like this. I know it's more work to set up, and certainly doing multiple 1d100 lists is out of the question, but put some more variability in there. That also lets you do things like have more extreme options weighted less...so maybe a lot of options on the list have 2 or 3 chances in 20 to come up, but the Cosmic Devourer Pokes His Proboscis Through the Multiverse and Nibbles your Cape only has a 1 in 20 chance. That makes it more rare, more special, and you can feel a little freer to give it consequences.
Fourth, define things. Example - Right now the result where a gate to another world appears and sucks people into it? We don't know how to determine where that gate appears. It's pretty crucial information to have. I assume the GM decides, but if so then say it because otherwise there'll be some bad blood.
Fifth, you might want to scale some of the effects with caster level. 4d10 is an incredibly huge amount of damage at 1st level. By 10th it's kind of meh. At 20 you barely notice it. Again, more power = more risk. When a level 20 Source blows his stack, people should probably be looking around for Mount Arrarat, because it was there when they left the cabin this morning. Conversely, a level 1 Source going nuclear should move a commensurately smaller molehill.
Not to say that a Surge is ever 'ignorable,' but scaling effects will really reduce the potential for one bad roll to cause a wipe...and that's something virtually guaranteed to create an OOC problem at any table.